Some constructive roads to Tychonoff
|
|
- Corey Hunt
- 6 years ago
- Views:
Transcription
1 Some constructive roads to Tychonoff Steven Vickers School of Computer Science, University of Birmingham, Birmingham, B15 2TT, UK. CLARENDON PRESS. OXFORD 2004
2 iv
3 INTRODUCTION v Abstract The Tychonoff Theorem is discussed with respect to point-free topology, from the point of view of both topos-valid and predicative mathematics. A new proof is given of the infinitary Tychonoff Theorem using predicative, choice-free methods for possibly undecidable index set. It yields a complete description of the finite basic covers of the product. 0.1 Introduction The Tychonoff theorem says that a product of compact topological spaces is still compact. (I do not assume Hausdorff separation here. By compact I mean having the finite subcover property, as in the Heine-Borel theorem; Bourbaki calls this quasicompact.) For finitary products this is fairly elementary. Surprisingly, the result extends to infinitary products, but there is a price the axiom of choice has to be assumed. In fact, as is well known, the infinitary Tychonoff theorem is equivalent to the axiom of choice [Kel50]. This is sometimes offered as a reason for using the axiom of choice: with it, one can prove many genuinely useful results such as infinitary Tychonoff. There are more that don t rely on the full power of the axiom of choice, but still need classical principles. These include the Heine-Borel theorem (that any bounded closed interval in the real line is compact) [FG82] and the Hofmann-Mislove theorem (that for a sober space there is a bijection between compact saturated subspaces and Scott open filters in the topology) [HM81], [Vic89]. The suggestion is that the theory of topology is unavoidably impoverished if it cannot call on all the classical reasoning principles. It therefore comes as a surprise to discover that in point-free topology, such theorems can often be proved constructively, at least if stated correctly. In this paper we illustrate this with the Tychonoff theorem. Its validity is quite unequivocal, with no special assumptions. For instance, we do not have to assume excluded middle either, and the indexing set does not have to have decidable equality. Neither do we have to use impredicative constructions. Tychonoff is actually a robust part of constructive topology. The problem lies not in forsaking choice, but in insisting on a point-set formulation of topology. Three things are jointly incompatible: Tychonoff, constructivity, and point-set topology. If we wish to keep Tychonoff then we must drop one of the others. However, it does not have to be the constructivity. The aim of this paper is to describe how it works if we decide to drop the point-set formulation Point-free topology In the usual point-set topology, a topological space is a set equipped with additional topological structure that can be axiomatized in various ways, classically equivalent to each other. Our preferred form here is via the open sets, defining a topology to be a family of subsets of the set of points, closed under finite intersection and arbitrary union. By contrast I shall use the phrase point-free topology as a generic label for approaches that do not start by assuming a set of points of the space. In these
4 vi approaches, a topological space cannot in general be described as a set of points equipped with extra structure. In practice, points are not excluded from the point-free discussion. Normally they are at least helpful for keeping in touch with topological intuitions, and there are tricks of categorical logic by which rigorous arguments can be conducted in terms of points. However, we cannot assume that the totality of all points can be collected together as a set. A space is something more general than a set. What is more, when a set of points can be extracted, we cannot assume that it adequately represents the collection of all points. There are two main versions of point-free topology that I shall consider, and they are genuinely different. They amount to the elaboration of point-free ideas in two radically different foundational settings, namely topos theory and predicative type theory. (I know more about the topos theory side, so you must excuse me if sometimes my knowledge of type theory is deficient.) The first version, used in topos theory, is locales [Joh82]. I shall explain those first because in many ways it helps to clarify what the second version is achieving. The second version, used in predicative type theory, is formal topology [Sam87] Locales Locale theory is based on the most direct interpretation of the phrase pointfree topology. In point-set topology, the topology is the collection of open sets. The idea is to use this in a point-free way by treating it as an abstract lattice, forgetting that it was ever a set of subsets of some set of points. The standard introduction is [Joh82] (or see also [Vic89]), and we give just a brief overview. Definition 0.1 A frame is a complete lattice in which binary meet distributes over arbitrary join. A frame homomorphism is a function between frames that preserves arbitrary joins and finite meets. (Naturally, the arbitrary joins and finite meets here correspond to the arbitrary unions and finite intersections of open sets.) We write Fr for the category of frames and frame homomorphisms, and Loc for its opposite the category of locales and (continuous) maps. By this definition a locale is just a frame, but it is best to keep them notationally distinct since the language for locales is quite different from that for frames. For instance, products of locales are coproducts of frames; sublocales are quotient frames. Locales are frames pretending to be topological spaces, and it is best to keep up the pretence. If X is a locale we shall write ΩX for the corresponding frame; and if f : X Y then we write f : ΩY ΩX for the corresponding frame homomorphism. Locales, then, are the spaces in this version of point-free topology. The points of a space X should be the maps from the one-point space 1 to X, and we can implement this with locales. The (discrete) topology on 1 is
5 INTRODUCTION vii the powerset P1. In the internal language of toposes P1 is just the subobject classifier Ω, and we shall write it as such. The best way to think of Ω is as the set of truthvalues, which classically is {true, false}. A point x of a locale X is then a frame homomorphism x : ΩX Ω. Now any function to Ω is equivalently described by its true kernel, the inverse image of {true}, and the function x is a frame homomorphism iff its true kernel is a completely prime filter, an upper set that is closed under finite meets and inaccessible by arbitrary joins. (Note that some of the other common characterizations of point ( [Joh82], [Vic89]) are not constructively equivalent. Some of this is because in the absence of excluded middle the homomorphism x is not determined by its false kernel (x ) 1 {false}.) In topos-valid mathematics, we can construct the set pt(x) of points of X. However, as we shall see it may be defective. Even classically there may fail to be enough points to distinguish between the opens, so that ΩX does not embed in P pt(x). There are two mismatches between locales and topological spaces. The first is that locales are intrinsically sober. This just means that the space contains all the points that can be reconstructed from the topology, and so must be a feature of any point-free approach. A non-sober topological space may have distinct elements of the point-set that cannot be distinguished topologically because they are in the same opens. In other words, the space may fail to be T 0. A non-sober space may also lack points such as the directed joins (with respect to the specialization order) present in every sober space. For example, a poset with its Alexandrov topology (the opens are the upper sets) is not sober in general, and to make it so you have to add directed joins by going to the ideal completion. That first mismatch is perhaps not so serious. One can argue that all decent spaces should be sober. Alternatively, you can express the duplications and omissions of points in a non-sober space by a map from a discrete locale X (ΩX = PX) to another locale. (These are the topological systems of [Vic89].) The second mismatch is more fundamental, and that is that locales do not always have enough points they may fail to be spatial. Indeed, some non-trivial locales fail to have any points at all. The technical manifestation of this is that the frame homomorphism ΩX P pt(x) fails to be 1-1. In classical mathematics the non-spatial locales are generally pathological, since the axiom of choice can be used to show the existence of enough points for wide classes of useful locales. Constructively, however, even necessary locales need not be spatial. A good example is the real line. The localic real line for which good mathematics holds for instance, the Heine-Borel theorem is the one presented as L(R) in [Joh82, IV.1.1]. Constructively this can easily be non-spatial [FH79]. Such non-spatiality may seem pathological wherever it occurs. After all, what kind of topological structure can it be that is not supported by the points? However, the well-known topological theorems work better with the non-spatial locales the purpose of this paper is to illustrate this for Tychonoff, and Heine- Borel has also been mentioned. Here s one way to imagine how the points might
6 viii not be the whole story. Often we are interested in other pieces within the space, for example line elements (maps from [0, 1]). Say a generalized point of X is a map from some domain Y (the stage of definition ) to X. Even if there are insufficient global points (stage of definition is 1), there are still plenty of generalized points. In fact, this comes rather cheaply, since the generic point (the identity map from X to itself) is enough for most purposes. It was understood quite early (see e.g. [JT84]) that locale theory is constructive in the topos-valid sense. The notion of internal frame can be defined in any topos, and the constructions one needs (for example, coproduct of frames for product of locales) can be carried through. Moreover, there is an extremely important relativization principle. Suppose X is a locale. There is a topos associated with it, namely its topos of sheaves which I shall write SX. What are the internal frames in SX? It turns out they are equivalent to continuous maps (of locales) with codomain X. Hence a constructive result about locales, interpreted in SX, can be turned into a result about maps into X in other words, generalized points of X. Thus the topos-valid constructivist discipline delivers a payoff. It is not merely a claim to moral superiority. One benefit is that constructive arguments about points can be applied also to the generalized points (as points at another stage of definition ). The sufficiency of these can therefore validate spatial reasoning about point-free topologies. This is exploited in [Vic99], [Vic04a], which also explain why the more stringent geometric constructivism is needed to ensure that the arguments can be transferred from one stage of definition to another. Topos-valid constructivism is completely choice-free. In general it is not even possible to choose one element out of two. Consider for example the topos of sheaves over the circle O. If the circle is represented as the complex numbers of unit modulus, then the squaring function z z 2, the Möbius double cover of the circle, is a local homeomorphism and hence equivalent to a sheaf. In the internal language of the topos SO, it is a set X, finite with decidable equality, satisfying but with no element 1 X Formal topologies x, y X. (x y z X. (z = x z = y)) Despite the success of locales in topos-valid mathematics, its use of impredicative constructions troubles some constructivists. These are constructions that presuppose a collection that already includes what one is trying to construct. The question often arises in connection with powersets PX, since if one is trying to construct some subset of X it would be impredicative to presuppose that PX the set of all subsets including the one being constructed is already to hand. In general predicative mathematics would not admit PX as a set. Unfortunately, many of the constructions of locale theory are impredicative. This includes the construction of pt(x), though I have already argued that we
7 INTRODUCTION ix may be able to do without it. More seriously, however, the frames themselves are impredicative. This is most obvious for discrete locales, whose frames are powersets. Then other frames such as ΩL(R) mentioned above, require impredicative constructions. And the very definition of frame, as complete lattice, describes joins in A by a function from PA to A. One way to understand formal topology is that it is obliged to do locale theory without the frames. In fact, techniques for this are already present in the ordinary practice of locale theory. The impredicative step is normally in ensuring that all joins (of subsets) are present. But it is often enough to work with a base of the topology, so that every open is a join of basic opens, and the base can often be constructed predicatively. More generally one might use a subbase, so that every open is a join of finite meets of subbasics. However, this makes no difference to the predicativity, since the finite powerset construction is inductive. (As elsewhere in this paper, finite means Kuratowski finite. The finite powerset FX can also be represented algebraically as the free semilattice over X.) In locale theory this use of bases or subbases appears in algebraic form, as presenting a frame by generators and relations. In [Joh82] it underlies the construction of ΩL(R). This is generated by basics (p, q) (p Q { }, q Q { }) subject to relations 1 = (, ) (p, q) (p, q ) = (max(p, p ), min(q, q )) (p, q) 0 if p q (p, s) (p, r) (q, s) if p q < r s (p, q) {(p, q ) p < p < q < q} if p < q The presentation itself is predicative: the set of generators, the set of relations and the sets of disjuncts in infinitary joins are all constructed predicatively. (This idea is explored in great detail in [Vic04a].) Hence, within predicative mathematics, the presentation can be used as a surrogate for the frame. This, roughly speaking, is what a formal topology is. More precisely, this is an inductively generated formal topology. One sees many different definitions in formal topology. To give some shape to the issues involved, I mention three different modes of variation. 1. There are different kinds of structure that can be interpreted as generators and relations. The different forms of structure tend to come out as different definitions of formal topology. For example, the site as described in [Joh82, II.2.11] provides one particular form of generators and relations. The generators are required to form a meet semilattice, and there are implied relations to say that the semilattice meet is preserved in the frame. 2. It used to be customary in formal topology to require spaces to be open in the sense of [JT84], namely that the unique map to 1 should be an open map. (Classically, all locales are open. But constructively it becomes an important
8 x issue.) For this a positivity predicate is needed on the basics in order to say in a positive way which are non-empty [Joh84], [Neg02]. A formal topology without positivity predicate is often called a formal cover. 3. The original definitions of formal topology required a specification of the full cover relation, i.e. to say for each set U of generators which generators a were to be less than U. Of course, is not itself a predicative set. But the information amounts to describing how proofs of a U may be constructed. More recently [CSSV03] showed how to use an axiom set, effectively a set of relations, to generate the full cover relation. Such a structure is called an inductively generated formal cover. Not all formal covers can be inductively generated. In what follows, we shall use the following definition of inductively generated formal cover. Definition 0.2 A flat site is a structure (P,, 0 ) where (P, ) is a preorder (i.e. transitive and reflexive), and 0 P PP has the following flat stability property: if a 0 U and b a, then there is some V b U such that b 0 V. (For subsets or elements U and V, we write U for the down closure of U with respect to and U V for U V.) The reason for calling this flat is as follows. In category theory there is a notion of flat functor from C to Set such if C is cartesian (has all finite limits) then flatness is equivalent to the functor being cartesian (preserves finite limits). See, e.g., [MLM92]. (This is also related to the notion of flat module in ring theory, using the idea from enriched category theory that a functor from C to Set can be considered a kind of module over C.) The notion of ordinary site [Joh82] is essentially a special case of our flat site in which P is a meet semilattice, i.e. a cartesian poset. In categorical logic, points of the corresponding locale can be understood as certain cartesian functors from P to Set, and in the flat site these generalize to the flat functors from P to Set. Definition 0.2 is just a rephrasing of the localized axioms sets of [CSSV03]. Their axiom set is an indexed family I(a) set [a : P ] together with a family of subsets C(a, i) P [a : P, i : I(a)]. Localized means that for any a c and i I(c), there exists j I(a) such that C(a, j) a C(c, i). Then our 0 comprises the instances of a 0 C(a, i). The full formal cover is generated from this by rules a U a U (reflexivity) a b b U a U ( -left) a 0 V V U a U (infinity) The flat site gives rise to a frame presentation in which the generators are the elements of P, and the relations are:
9 COMPACTNESS xi 0.2 Compactness 1 P a b = {c c a, c b} a U (a 0 U) The notion of compactness translates easily from spaces to locales. A locale X is compact if the top open 1 ΩX has the property that every cover has a finite subcover. Alternatively, if a directed subset S has its join equal to 1, then 1 must already be in S i.e. {1} is Scott open. This is straightforward, but notice that compactness of the locale and its space of points pt(x) become two unrelated properties. Let Ω pt(x) be the topology induced on pt(x), the image of the frame homomorphism ΩX P pt(x), and let us write F for the filter of ΩX comprising those opens that map to top in Ω pt(x). If F is not spatial, so ΩX Ω pt(x) is not 1-1, then F may be different from {1}. Compactness for pt(x) is equivalent to saying that F is Scott open if a directed join S is in F, then S already has an element in F. Scott openness of F neither implies nor is implied by Scott openness of {1}. In fact, this explains something of the gap between spatial Tychonoff (requiring choice in general) and localic Tychonoff (no choice needed). Even for spatial locales, the product need not be spatial. Hence compactness of the product locale does not imply compactness of the product space the two questions are separate. Despite the simplicity of the definition of localic compactness, in practice it is a non-trivial question. This is because it is rare for the frame structure to be given explicitly in a concrete form. For instance, if the frame is presented by basic generators and relations it is not in general clear when one open is covered by a family of basics. The impredicative definition of the cover relation in effect the least frame congruence containing the relations is little help. In particular this is a problem with a product i X i of locales, whose coproduct frame is most easily presented by a disjoint union of presentations for the frames ΩX i. This coproduct frame may also be described as a tensor product of complete join semilattices, but that is no real help here because just as with linear tensor products the elements cannot be expressed in any canonical form. A direct predicative approach requires some knowledge of the full cover relation. Let us outline some sharper approaches to the question Preframes A preframe is a poset with finite meets and directed joins, with meet distributing over directed joins. A preframe homomorphism preserves finite meets and directed joins. The importance of preframes lies in the fact that for a subset F of a frame ΩX, F is a Scott open filter iff its characteristic function to Ω is a preframe
10 xii homomorphism filteredness and Scott openness correspond to preservation of finite meets and directed joins respectively. A simple proof of Tychonoff using preframe techniques was given in [JV91]. This can be expressed neatly within locale theory by the upper powerlocale P U X (see [Vic97], and also [Vic04a]). By definition its frame ΩP U (X) is generated as frame by the elements of ΩX, respecting the preframe structure of ΩX. Hence maps from Y to P U (X) are equivalent to preframe homomorphisms from ΩX to ΩY and the points of P U X are the Scott open filters of ΩX. Johnstone s localic version of the Hofmann-Mislove Theorem see [Vic97], deriving from [Joh85] says that the Scott open filters of ΩX correspond to compact fitted sublocales of X, where a sublocale is fitted if it is a meet of open sublocales. (Classically this corresponds to subspaces that are saturated, i.e. upper closed under the specialization order.) Hence P U X is indeed a powerlocale, its points being certain sublocales of X. The correspondence is order reversing, and a bottom point of P U X corresponds to the greatest possible compact sublocale of X, namely X itself. [Vic95] shows that proving compactness of X is equivalent to finding a bottom point of P U (X), in the strong sense that the composite!; : P U X 1 P U X is less than the identity map in the specialization order. This condition says that is not just least amongst the global points 1 P U X. It is also less than the generic point id : P U X P U X, and this makes it least amongst all generalized points Y P U X. All this is impredicative, but it can be made predicative. The preframe coverage theorem of [JV91] shows how to convert presentations of frames by generators and relations into preframe presentations of the same frames, and so shows how to convert frame presentations of ΩX into frame presentations of ΩP U X. This can be made into a predicative construction within formal topology. Proposition 0.3 Let (P,, 0 ) be a flat site presenting locale X. Then P U X is presented by the flat site (FP, L, 0 ) where FP is the (Kuratowski) finite powerset of P, L is the lower order on FP, defined by S L T iff for every x S there is some y T with x y, and 0 is given by the following. Suppose a i 0 U i (1 i n). Let A = {a i 1 i n}. Then for every S FP, A S 0 {T S T L n i=1 U i }. Proof We merely sketch the proof here. ΩX is presented as frame by generators P and relations as given after Definition 0.2. We can write it as Fr P (qua preorder) 1 P a b (a b) a U (if a 0 U).
11 COMPACTNESS xiii where qua preorder indicates extra implicit relations to say the preorder structure of P is preserved in the frame. The free join semilattice over the preorder P is FP/ L, with join represented by union. We can therefore transform the presentation into an equivalent one, Fr FP/ L (qua -semilattice) 1 FP ({a} S) ({b} S) {(T S) T L {a}, T L {b}} ({a} S) {(T S) T L U} (if a 0 U, S FP ). ( indicates a join that is known to be directed. We have not distinguished between on the one hand the elements of FP/ L, equivalence classes with respect to the equivalence relation corresponding to L, and on the other hand the elements of FP that represent them.) Here the relations have been put in the join stable form required for the preframe coverage theorem [JV91], and from that we find that ΩP U X can be given exactly the same presentation, except that qua -semilattice is replaced by qua preorder under L. Using some induction on the finite sets, we find that the middle relation scheme is equivalent to (A) (B) {(T ) T L A, T L B} for A, B FP. Thus those first two relation schemes are equivalent to the implicit relations in a flat site on (FP, L ). The final relation scheme does not satisfy the flat stability condition, but it is equivalent to the relation scheme given in the statement of the theorem, which does. It follows that the upper powerlocale can also be accessed in predicative theories. The same compactness criterion existence of a suitable point of the powerlocale can then be expressed. The argument given so far for its correctness has used the impredicative results about preframes, but [Vic02] gives a direct predicative proof. After a little simplification, it appears there as Theorem 0.4 Let (P,, 0 ) be a flat site presenting a locale X. Then X is compact iff there is a subset F of FP such that 1. F is upper closed with respect to L. 2. F is inhabited. 3. If a 0 U and {a} T F, then U 0 T F for some U 0 FU. 4. If S F then P S (i.e. g P. g S). In that case, F is necessarily the set of all finite basic covers of X. We note briefly that it is not only preframe homomorphisms that can be captured predicatively in this way. So too can arbitrary dcpo morphisms (Scott
12 xiv continuous functions) between frames. (A dcpo is a directed complete poset, i.e. a poset with all directed joins.) This is done using the double powerlocale PX ( [JV91], [Vic04a]), for which the frame ΩPX is the free frame generated by ΩX and preserving its dcpo structure. The maps from Y to PX are equivalent to Scott continuous functions from ΩX to ΩY, and PX can be constructed by predicative constructions on presentations ( [Vic04a], [VT04]). Thus the double powerlocale can also be defined on inductively generated formal topologies. 0.3 Tychonoff We can now illustrate the techniques with a proof of Tychonoff that is valid both in topos theory and in predicative mathematics. It assumes neither finiteness nor decidability of equality for the indexing set for the locales of which the product is taken. In topos-valid locale theory this result appears to be due to Vermeulen [Ver86]. In formal topology, following an earlier treatment of [Coq92], the infinitary Tychonoff was proved in [NV97] without choice but under the assumption that the indexing set had decidable equality. This arose from the way that basic opens for the product i X i finite meets of opens taken from the components were normalized into elements of i ΩX i in which all but finitely many components are 1. This normalization can only be done effectively if there is decidable equality for indexes. Subsequently, [Coq03] gave a simple choice-free predicative proof without decidable equality. His argument rests on the fact that for any spectral locale X, there is a least compact sublocale Y whose fitted hull (= saturation) is the whole of X. It follows that every sublocale between Y and X is compact. Coquand shows how to describe a product locale in this way. (Coquand has remarked separately that the underlying construction is a localic version of the maximal spectrum described spatially in [Joh82, II.3.5].) Though elegant, Coquand s proof requires some preparation before it can be put into effect. It relies on having each locale presented using a distributive lattice of generators for which the order coincides with the order in the presented frame, and getting that is non-trivial. We now give a proof that shows how from general flat sites, the finite covers of the product can be calculated. Proposition 0.5 Let (P i,, 0 ) be a flat site for each i I. Then the product of the corresponding locales is presented by a flat site (P,, 0 ) defined as follows. First, let (P, ) = i I P i be the poset coproduct. As a set it is the disjoint union, {(i, x) i I, x P i }, with (i, x) (j, y) iff i = j and x y in P i. Its elements are subbasics. Now define P = FP ordered by U, i.e. A B iff b B. a A. a b. Its elements represent finite meets of subbasics. P/ U is in fact the free meet semilattice over the poset P, meet being represented by union. Covers are defined as follows: 1. If i I and B P, then
13 TYCHONOFF xv 2. If i I, a, a P i and B P, then B 0 {{(i, a)} B a P i }. {(i, a), (i, a )} B 0 {{(i, c)} B c a, c a }. 3. If i I, a 0 U in P i, and B P, then {(i, a)} B 0 {{(i, u)} B u U}. Proof First note that this is indeed a flat site; in fact it is an ordinary site. (P is a meet semilattice and the coverage has meet stability.) The frame for the product is presented by putting together the presentations for the original frames. For clarity, let us write α i for the injections of generators. Then the frame is presented as Fr α i (a) (i I, a P i ) This is isomorphic to α i (a) α i (a ) (i I, a a in P i ) 1 a P i α i (a) (i I) α i (a) α i (a ) {α i (c) c a, c a } (i I, a, a P i ) Fr P (qua = semilattice) α i (a) u U α i(u) (i I, a 0 U in P i ). B a P i ({(i, a)} B) (i I, B P ) {(i, a), (i, a )} B {{(i, c)} B c a, c a } (i I, a, a P i, B P ) {(i, a)} B u U {(i, u)} B (i I, a 0 U in P i, B P ). The qua notation denotes additional relations to preserve the -semillatice structure (concretely ) of P. In one direction, the isomorphism takes α i (a) ({(i, a)}), while in the other it takes A {α i (a) (i, a) A}. To say this predicatively, we are describing two mutually inverse continuous maps between the corresponding formal topologies. This second presentation corresponds to the product site described in the statement. To prove Tychonoff, we use Theorem 0.4. Suppose we have flat sites (P i,, 0 ) and are given sets F i describing compactness for the P i s. We show how to construct a corresponding set F for the product. The main point of interest is that F itself can be defined without reference to the full coverage. The full coverage and its inductive generation need to be considered only when showing that every set in F covers the product space; but this is hardly surprising, because the corresponding facts for the F i s were described in terms of.
14 xvi We must find a way to characterize the finite basic covers. We give an informal argument as motivation; applying the theorem will confirm its correctness. Each subbasic (i, a) in P is of the form (spatially) {(x j ) j I x i a}: think of this as a product of a (at i) P j (everywhere else). A basic in P is a set of these representing a meet, and that can be thought of as a product of specified a s at finitely many specified i s, times P j everywhere else. (However, we must also allow for the fact that some i may occur more than once.) We want to know when a join of these meets covers the entire product, and the trick is to use distributivity to change it to a meet of joins. Then every one of the joins must be the whole product. By distributivity, A = Im γ A A γ Ch(A) where Ch(A) is the set of choices of A, i.e. [Vic04b] the finite total relations γ from A to A such that if (A, a) γ then a A, and Im γ is the image of γ (under the second projection from A A to A). Now consider a finite join of subbasics B. This is (it will turn out) a cover of the entire product iff at some i its components cover P i. Classically one sees this as follows. Suppose at every i we have some point x i that is not in any subbasic b in B. Then the point (x i ) i I is not in B. Hence (classically) if B does cover the product, then there is some i and some finite cover S of P i (so S F i ) such that {i} S B. This idea lies behind our definition of F in the Theorem. Theorem 0.6. (Infinitary Tychonoff) Let (P i,, 0 ) (i I) be flat sites for compact spaces, equipped with sets F i FP i satisfying the conditions of Theorem 0.4. Let the product site P be defined as above. Let F FP be defined such that A F iff for every γ Ch(A) there is some i and some S F i such that for every a S we have (i, a) Im(γ). Then F satisfies the conditions of Theorem 0.4 for P, and hence shows that P is compact. Proof In the definition of F, we should like to say that for some i, Im γ covers P i : or {a (i, a) Im γ} F i. But we have to be somewhat careful, since if I does not have decidable equality then {a (i, a) Im γ} need not be finite. Nonetheless, let us abuse language and say Im γ covers P i. Note also that if some F i contains, so that P i gives an empty locale and so does the whole product, then every A is in F. We verify the four conditions in Theorem 0.4. Condition 1, F is upper closed with respect to L. Suppose A F and A L B. Let δ Ch(B). If A A then A U B for some B B. There is some b B Im δ, and a b for some a A. In short, A A. a A. b Im δ. a b. It follows that there is some γ Ch(A) such that Im γ L Im δ. Now because A F we deduce that Im γ covers some P i, and it follows that Im δ covers the same P i so B F.
15 TYCHONOFF xvii Condition 2, F is inhabited. { } is vacuously in F, because it has no choices. Condition 3. There are three parts to check, corresponding to the three axiom schemes in Proposition 0.5. We use Lemma 0.7, which is proved separately. For scheme 1, we need that if i I, B P and {B} C F, then there is some S FP i such that {{(i, a)} B a S } C F. In the lemma, take S = and φ the whole of P i. Given T, there is some S such that S T covers P i ; just choose S to be any element of F i (which is inhabited by hypothesis). For scheme 2, if i I, a 1, a 2 P i, B P and {{(i, a 1 ), (i, a 2 )} B} C F, then there is some S F(a 1 a 2 ) such that {{(i, a)} B a S } C F. Here S = {a 1, a 2 } and φ = a 1 a 2. If {a 1 } T and {a 2 } T both cover P i then so does ({a 1 } T ) ({a 2 } T ) and hence so does some finite subset. This enables us to find S. For scheme 3, if i I, a 0 U in P i, B P and {{(i, a)} B} C F, then there is some U 0 FU such that {{(i, u)} B u U 0 } C F. Here S = {a} and φ = U. If {a} T F i then by hypothesis U 0 T F i for some U 0 FU. Condition 4, if A F then P A. Let us write B = {Im γ γ Ch(A)}. For every γ Ch(A) we can find S in some F i such that {i} S Im γ, and it follows that we can find B FP such that every B in B is {i} S for some i and S F i ; every B in B is included in some Im γ in B ; every Im γ in B includes some B in B. The last two imply that B U B and B L B (recalling that the order used on P = FP is U, which includes ). Now let C = {Im δ δ Ch(B)}. We show (i) C L A, and (ii) ( ) C, and these together imply that ( ) A. For the first, take δ Ch(B). For every γ Ch(A) we have that Im γ includes some B B and so meets Im δ. By the Diagonalization Lemma of [Vic04b] it follows that A Im δ for some A A. (Classically, if no A A is included in Im δ then there is a choice that avoids Im δ. But with these finite sets there is a constructive proof.) For the second, we use induction on B. If B =, then it has only one choice, which is empty, and so C = { }. Now suppose it holds for B 0 ; we prove it for B = B 0 {{i} S} where S F i. We have By induction, C {Im δ {(i, b)} δ Ch(B 0 ), b S}. ( ) {Im δ δ Ch(B 0 )}.
16 xviii By definition of 0 for P, Im δ 0 {Im δ {(i, a)} a P i }. Since S F i, we have a S for each a P i, and a straightforward induction on the proof of a S then shows that We can now use transitivity of. Im δ {(i, a)} {Im δ {(i, b)} b S}. Now here is the lemma that was promised in proving condition 3. Lemma 0.7 Under the hypotheses of Theorem 0.6, suppose we have i I, S FP i and φ P i with the property that, for every T FP i, if a S. {a} T F i then there is some S FP i with S φ and S T F i. Then if B P, C FP and {({i} S) B} C F, there is some S FP i with S φ and {{(i, a)} B a S } C F. Proof Suppose B and C are given. Now suppose γ Ch(C). For every a S we have {(i, a)} Im γ covers some P j, so there is some S F j such that {j} S {(i, a)} Im γ. We can deduce that either Im γ covers some P j, or {(i, a)} Im γ covers P i. We can therefore decompose Ch(C) as a union of finite sets, D D, such that if γ D then Im γ covers some P j, and if γ D then {(i, a)} Im γ covers P i for every a S. If γ D then for each a S we can find T P i such that {a} T F i and {i} T Im γ, and by taking their union we can assume that a single T does for all the a s. Then we can find S with S φ and S T F i. By taking the union of the S s we can assume that a single S such that ({i} S ) Im γ covers P i for all γ D. We now show that {{(i, a)} B a S } C F. For any choice of {{(i, a)} B a S } C, its image contains a set of the form Im γ Im δ, where γ Ch(C) and δ Ch({{(i, a)} B a S }). For each a S, we have either Im δ meets B or (i, a) Im δ. If the former holds for some a, then there is some choice of {({i} S) B} C whose image is a subset of Im γ Im δ, and from {({i} S) B} C F we deduce that Im γ Im δ covers some P j. Alternatively, suppose {i} S Im δ. It suffices then to know that ({i} S ) Im γ covers some P j. If γ D then Im γ covers some P j, while if γ D then by construction of S we have ({i} S ) Im γ covers P i. In either case we are done. 0.4 Synthetic locale theory We conclude with some remarks on an approach that promises to lay bare many issues of topology, both point-set and point-free. This is the synthetic topology of Escardó [Esc04]. It uses the lambda calculus to express maps, and the Tychonoff theorem (at least, binary Tychonoff) is a good illustration. Recall that X is compact iff there is a Scott continuous map! : ΩX Ω that is right adjoint to the unique frame homomorphism. For sober spaces, all
17 SYNTHETIC LOCALE THEORY xix continuous maps are Scott continuous (with respect to the specialization order). Also, the opens of X are equivalent to continuous maps from X to the Sierpinski space S, so we can identify ΩX with the function space S X. Hence, compactness of X can be expressed by a map X : S X S right adjoint to the map S! : S S X. We can think of the points of S as being truth values, the top (open) point being true, and then (a) is the truth value of a = X. If X and Y are both compact, then the corresponding map for X Y can be expressed very easily as X Y (u) = Y (λy. X (λx. u(x, y))). To put it another way, X Y : S X Y S is the composite =; ( X ) Y ; Y : S X Y = (S X ) Y S Y S. This is the required right adjoint, and if everything preserves continuity then we get the required Scott continuity. But there s an obvious flaw in the argument! The function space S X only exists if X is locally compact (this holds for locales as well as spaces). Apparently, it proves Tychonoff only for locally compact (and sober) spaces. However, it is possible to get round this by embedding one s category of spaces in a larger category in which the exponentials exist. Escardó refers to real spaces (in the original category) and complex spaces (in the supercategory). It then remains only to show that morphisms between the complex function spaces do indeed give the required Scott continuous functions between frames. For locales, the requisite results have been proved in [VT04]. There the category Loc of locales is embedded (by the Yoneda embedding) in the category Set Locop of presheaves over Loc. The fundamental lemma then is that presheaf morphisms (natural transformations) from S X to S Y correspond to Scott continuous functions from ΩX to ΩY. This allows us to use the above construction of X Y as a proof of binary Tychonoff. Despite the set-theoretic difficulties, it is to be hoped that a predicative argument can also be found to justify such synthetic methods in formal topology. The infinitary Tychonoff theorem is less well understood from this point of view, but it seems to play the role of a termination principle for recursive algorithms.
18 REFERENCES [Coq92] T. Coquand, An intuitionistic proof of Tychonoff s theorem, Journal of Symbolic Logic 57 (1992), [Coq03] T. Coquand, Compact spaces and distributive lattices, Journal of Pure and Applied Algebra 184 (2003), 1 6. [CSSV03] T. Coquand, G. Sambin, J. Smith, and S. Valentini, Inductively generated formal topologies, Annals of Pure and Applied Logic 124 (2003), [Esc04] Martín Escardó, Synthetic topology of data types and classical spaces, Bellairs Workshop: Domain Theoretic Methods in Probabilistic Processes (P. Panangaden and J. Desharnais, eds.), Electronic Notes in Theoretical Computer Science, vol. 87, Elsevier, [FG82] M.P. Fourman and R.J. Grayson, Formal spaces, The L.E.J. Brouwer Centenary Symposium (Troelstra and van Dalen, eds.), North Holland, 1982, pp [FH79] M.P. Fourman and J.M.E. Hyland, Sheaf models for analysis, Applications of Sheaves (M.P. Fourman, C.J. Mulvey, and D. Scott, eds.), Lecture Notes in Mathematics, no. 753, Springer, 1979, pp [HM81] Karl H. Hofmann and Michael W. Mislove, Local compactness and continuous lattices, Continuous Lattices: Proceedings, Bremen, 1979 (B. Banaschewski and R.-E. Hoffmann, eds.), Lecture Notes in Mathematics, no. 871, Springer, 1981, pp [Joh82] P.T. Johnstone, Stone spaces, Cambridge Studies in Advanced Mathematics, no. 3, Cambridge University Press, [Joh84] P.T. Johnstone, Open locales and exponentiation, Contemporary Mathematics 30 (1984), [Joh85] P.T. Johnstone, Vietoris locales and localic semi-lattices, Continuous Lattices and their Applications (R.-E. Hoffmann, ed.), Pure and Applied Mathematics, no. 101, Marcel Dekker, 1985, pp [JT84] A. Joyal and M. Tierney, An extension of the Galois theory of Grothendieck, Memoirs of the American Mathematical Society 309 (1984). [JV91] P.T. Johnstone and S.J. Vickers, Preframe presentations present, Category Theory Proceedings, Como 1990 (A. Carboni, M.C. Pedicchio, and G. Rosolini, eds.), Lecture Notes in Mathematics, no. 1488, Springer-Verlag, 1991, pp [Kel50] J.L. Kelley, The Tychonoff product theorem implies the axiom of choice, Fundamenta Mathematicae 37 (1950), [MLM92] S. Mac Lane and I. Moerdijk, Sheaves in geometry and logic, Springer Verlag, xx
19 REFERENCES xxi [Neg02] Sara Negri, Continuous domains as formal spaces, Mathematical Structures in Computer Science 12 (2002), [NV97] S. Negri and S Valentini, Tychonoff s theorem in the framework of formal topologies, Journal of Symbolic Logic 62(4) (1997), [Sam87] G. Sambin, Intuitionistic formal spaces a first communication, Mathematical Logic and its Applications (Skordev, ed.), Plenum, 1987, pp [Ver86] J.J.C. Vermeulen, Constructive techniques in functional analysis, Ph.D. thesis, University of Sussex, [Vic89] S.J. Vickers, Topology via logic, Cambridge University Press, [Vic95] S.J. Vickers, Locales are not pointless, Theory and Formal Methods of Computing 1994 (London) (C.L. Hankin, I.C. Mackie, and R. Nagarajan, eds.), Imperial College Press, 1995, pp [Vic97] S.J. Vickers, Constructive points of powerlocales, Math. Proc. Cam. Phil. Soc. 122 (1997), [Vic99] S.J. Vickers, Topical categories of domains, Mathematical Structures in Computer Science 9 (1999), [Vic02] Steven Vickers, Compactness in locales and formal topology, accepted for Proceedings of 2nd Workshop on Formal Topology, Venice Draft available on web at [Vic04a] S.J. Vickers, The double powerlocale and exponentiation: A case study in geometric reasoning, Theory and Applications of Categories 12 (2004), [Vic04b] Steven Vickers, Entailment systems for stably locally compact locales, Theoretical Computer Science 316 (2004), [VT04] S.J. Vickers and C.F. Townsend, A universal characterization of the double powerlocale, Theoretical Computer Science 316 (2004),
University of Oxford, Michaelis November 16, Categorical Semantics and Topos Theory Homotopy type theor
Categorical Semantics and Topos Theory Homotopy type theory Seminar University of Oxford, Michaelis 2011 November 16, 2011 References Johnstone, P.T.: Sketches of an Elephant. A Topos-Theory Compendium.
More informationTopos Theory. Lectures 17-20: The interpretation of logic in categories. Olivia Caramello. Topos Theory. Olivia Caramello.
logic s Lectures 17-20: logic in 2 / 40 logic s Interpreting first-order logic in In Logic, first-order s are a wide class of formal s used for talking about structures of any kind (where the restriction
More informationCosheaves and connectedness in formal topology
Cosheaves and connectedness in formal topology Steven Vickers School of Computer Science, University of Birmingham, Birmingham, B15 2TT, UK s.j.vickers@cs.bham.ac.uk July 25, 2013 Abstract The localic
More informationACLT: Algebra, Categories, Logic in Topology - Grothendieck's generalized topological spaces (toposes)
ACLT: Algebra, Categories, Logic in Topology - Grothendieck's generalized topological spaces (toposes) Steve Vickers CS Theory Group Birmingham 2. Theories and models Categorical approach to many-sorted
More informationBinary positivity in the language of locales
Binary positivity in the language of locales Francesco Ciraulo Department of Mathematics University of Padua 4 th Workshop on Formal Topology June 15-20 2012, Ljubljana Francesco Ciraulo (Padua) Binary
More informationSynthetic Computability
Synthetic Computability Andrej Bauer Department of Mathematics and Physics University of Ljubljana Slovenia MFPS XXIII, New Orleans, April 2007 What is synthetic mathematics? Suppose we want to study mathematical
More informationAdjunctions! Everywhere!
Adjunctions! Everywhere! Carnegie Mellon University Thursday 19 th September 2013 Clive Newstead Abstract What do free groups, existential quantifiers and Stone-Čech compactifications all have in common?
More informationTopology Proceedings. COPYRIGHT c by Topology Proceedings. All rights reserved.
Topology Proceedings Web: http://topology.auburn.edu/tp/ Mail: Topology Proceedings Department of Mathematics & Statistics Auburn University, Alabama 36849, USA E-mail: topolog@auburn.edu ISSN: 0146-4124
More informationBoolean Algebra and Propositional Logic
Boolean Algebra and Propositional Logic Takahiro Kato September 10, 2015 ABSTRACT. This article provides yet another characterization of Boolean algebras and, using this characterization, establishes a
More informationEmbedding locales and formal topologies into positive topologies
Embedding locales and formal topologies into positive topologies Francesco Ciraulo Giovanni Sambin Department of Mathematics, University of Padova Via Trieste 63, 35121 Padova, Italy ciraulo@math.unipd.it,
More informationLocalic completion of generalized metric spaces II: Powerlocales
1 48 ISSN 1759-9008 1 Localic completion of generalized metric spaces II: Powerlocales STEVEN VICKERS Abstract: The work investigates the powerlocales (lower, upper, Vietoris) of localic completions of
More informationTopos Theory. Lectures 21 and 22: Classifying toposes. Olivia Caramello. Topos Theory. Olivia Caramello. The notion of classifying topos
Lectures 21 and 22: toposes of 2 / 30 Toposes as mathematical universes of Recall that every Grothendieck topos E is an elementary topos. Thus, given the fact that arbitrary colimits exist in E, we can
More informationA bitopological point-free approach to compactifications
A bitopological point-free approach to compactifications Olaf Karl Klinke a, Achim Jung a, M. Andrew Moshier b a School of Computer Science University of Birmingham Birmingham, B15 2TT England b School
More informationBoolean Algebra and Propositional Logic
Boolean Algebra and Propositional Logic Takahiro Kato June 23, 2015 This article provides yet another characterization of Boolean algebras and, using this characterization, establishes a more direct connection
More informationSome glances at topos theory. Francis Borceux
Some glances at topos theory Francis Borceux Como, 2018 2 Francis Borceux francis.borceux@uclouvain.be Contents 1 Localic toposes 7 1.1 Sheaves on a topological space.................... 7 1.2 Sheaves
More informationAn adjoint construction for topological models of intuitionistic modal logic Extended abstract
An adjoint construction for topological models of intuitionistic modal logic Extended abstract M.J. Collinson, B.P. Hilken, D.E. Rydeheard April 2003 The purpose of this paper is to investigate topological
More informationIII A Functional Approach to General Topology
III A Functional Approach to General Topology Maria Manuel Clementino, Eraldo Giuli and Walter Tholen In this chapter we wish to present a categorical approach to fundamental concepts of General Topology,
More informationNotes about Filters. Samuel Mimram. December 6, 2012
Notes about Filters Samuel Mimram December 6, 2012 1 Filters and ultrafilters Definition 1. A filter F on a poset (L, ) is a subset of L which is upwardclosed and downward-directed (= is a filter-base):
More informationComparing cartesian closed categories of (core) compactly generated spaces
1 Comparing cartesian closed categories of (core) compactly generated spaces By MARTÍN ESCARDÓ School of Computer Science University of Birmingham, UK JIMMIE LAWSON Department of Mathematics Louisiana
More informationvia Topos Theory Olivia Caramello University of Cambridge The unification of Mathematics via Topos Theory Olivia Caramello
in University of Cambridge 2 / 23 in in In this lecture, whenever I use the word topos, I really mean Grothendieck topos. Recall that a Grothendieck topos can be seen as: a generalized space a mathematical
More informationElementary (ha-ha) Aspects of Topos Theory
Elementary (ha-ha) Aspects of Topos Theory Matt Booth June 3, 2016 Contents 1 Sheaves on topological spaces 1 1.1 Presheaves on spaces......................... 1 1.2 Digression on pointless topology..................
More informationA Note on Extensional PERs
A Note on Extensional PERs W. P. Stekelenburg March 2, 2010 Abstract In the paper Extensional PERs by P. Freyd, P. Mulry, G. Rosolini and D. Scott, a category C of pointed complete extensional PERs and
More informationA Discrete Duality Between Nonmonotonic Consequence Relations and Convex Geometries
A Discrete Duality Between Nonmonotonic Consequence Relations and Convex Geometries Johannes Marti and Riccardo Pinosio Draft from April 5, 2018 Abstract In this paper we present a duality between nonmonotonic
More informationA Non-Topological View of Dcpos as Convergence Spaces
A Non-Topological View of Dcpos as Convergence Spaces Reinhold Heckmann AbsInt Angewandte Informatik GmbH, Stuhlsatzenhausweg 69, D-66123 Saarbrücken, Germany e-mail: heckmann@absint.com Abstract The category
More informationLOCALES AND TOPOSES AS SPACES
Chapter 1 LOCALES AND TOPOSES AS SPACES Steven Vickers University of Birmingham 1. Introduction Second Reader Guram Bezhanishvili New Mexico State University Mac Lane and Moerdijk, 1992, in their thorough
More informationBoolean Algebras, Boolean Rings and Stone s Representation Theorem
Boolean Algebras, Boolean Rings and Stone s Representation Theorem Hongtaek Jung December 27, 2017 Abstract This is a part of a supplementary note for a Logic and Set Theory course. The main goal is to
More informationTopos Investigations of the Extended Priestley Duality. Kostas Viglas
Topos Investigations of the Extended Priestley Duality Kostas Viglas April 9, 2003 Abstract This is a brief summary of the original results in the thesis. First, there is a localic version of the correspondence
More informationSJÄLVSTÄNDIGA ARBETEN I MATEMATIK
SJÄLVSTÄNDIGA ARBETEN I MATEMATIK MATEMATISKA INSTITUTIONEN, STOCKHOLMS UNIVERSITET Equivariant Sheaves on Topological Categories av Johan Lindberg 2015 - No 7 MATEMATISKA INSTITUTIONEN, STOCKHOLMS UNIVERSITET,
More informationAn introduction to toposes. Richard Pettigrew Department of Philosophy University of Bristol
n introduction to toposes Richard Pettigrew Department of Philosophy University of Bristol Contents 1 Motivating category theory 1 1.1 The idea behind category theory.................. 1 2 The definition
More information14 Lecture 14: Basic generallities on adic spaces
14 Lecture 14: Basic generallities on adic spaces 14.1 Introduction The aim of this lecture and the next two is to address general adic spaces and their connection to rigid geometry. 14.2 Two open questions
More informationLecture 9: Sheaves. February 11, 2018
Lecture 9: Sheaves February 11, 2018 Recall that a category X is a topos if there exists an equivalence X Shv(C), where C is a small category (which can be assumed to admit finite limits) equipped with
More informationConstructive version of Boolean algebra
Constructive version of Boolean algebra Francesco Ciraulo, Maria Emilia Maietti, Paola Toto Abstract The notion of overlap algebra introduced by G. Sambin provides a constructive version of complete Boolean
More informationA Grothendieck site is a small category C equipped with a Grothendieck topology T. A Grothendieck topology T consists of a collection of subfunctors
Contents 5 Grothendieck topologies 1 6 Exactness properties 10 7 Geometric morphisms 17 8 Points and Boolean localization 22 5 Grothendieck topologies A Grothendieck site is a small category C equipped
More informationFrom Constructibility and Absoluteness to Computability and Domain Independence
From Constructibility and Absoluteness to Computability and Domain Independence Arnon Avron School of Computer Science Tel Aviv University, Tel Aviv 69978, Israel aa@math.tau.ac.il Abstract. Gödel s main
More informationSynthetic Computability (Computability Theory without Computers)
Synthetic Computability (Computability Theory without Computers) Andrej Bauer Department of Mathematics and Physics University of Ljubljana Slovenia Gargnano, Lago di Garda, September 2006 What is synthetic
More informationAbstract and Variable Sets in Category Theory 1
Abstract and Variable Sets in Category Theory 1 John L. Bell In 1895 Cantor gave a definitive formulation of the concept of set (menge), to wit, A collection to a whole of definite, well-differentiated
More informationThe equivalence axiom and univalent models of type theory.
The equivalence axiom and univalent models of type theory. (Talk at CMU on February 4, 2010) By Vladimir Voevodsky Abstract I will show how to define, in any type system with dependent sums, products and
More informationHOMOLOGY THEORIES INGRID STARKEY
HOMOLOGY THEORIES INGRID STARKEY Abstract. This paper will introduce the notion of homology for topological spaces and discuss its intuitive meaning. It will also describe a general method that is used
More informationAlgebraic Geometry
MIT OpenCourseWare http://ocw.mit.edu 18.726 Algebraic Geometry Spring 2009 For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms. 18.726: Algebraic Geometry
More informationhal , version 1-21 Oct 2009
ON SKOLEMISING ZERMELO S SET THEORY ALEXANDRE MIQUEL Abstract. We give a Skolemised presentation of Zermelo s set theory (with notations for comprehension, powerset, etc.) and show that this presentation
More informationGelfand spectra in Grothendieck toposes using geometric mathematics
Gelfand spectra in Grothendieck toposes using geometric mathematics Bas Spitters Formerly, Institute for Computing and Information Sciences, University of Nijmegen Steven Vickers bawspitters@gmail.com
More informationClosure operators on sets and algebraic lattices
Closure operators on sets and algebraic lattices Sergiu Rudeanu University of Bucharest Romania Closure operators are abundant in mathematics; here are a few examples. Given an algebraic structure, such
More informationOlivia Caramello. University of Insubria - Como. Deductive systems and. Grothendieck topologies. Olivia Caramello. Introduction.
duality University of Insubria - Como 2 / 27 duality Aim of the talk purpose of this talk is to illustrate the relevance of the notion of topology. I will show that the classical proof system of geometric
More informationFoundations of Mathematics MATH 220 FALL 2017 Lecture Notes
Foundations of Mathematics MATH 220 FALL 2017 Lecture Notes These notes form a brief summary of what has been covered during the lectures. All the definitions must be memorized and understood. Statements
More informationLecture 2: Syntax. January 24, 2018
Lecture 2: Syntax January 24, 2018 We now review the basic definitions of first-order logic in more detail. Recall that a language consists of a collection of symbols {P i }, each of which has some specified
More informationUnbounded quantifiers and strong axioms in topos theory
Unbounded quantifiers and in topos A. University of Chicago November 14, 2009 The motivating question What is the topos-theoretic counterpart of the strong set-theoretic axioms of Separation, Replacement,
More informationCATEGORY THEORY. Cats have been around for 70 years. Eilenberg + Mac Lane =. Cats are about building bridges between different parts of maths.
CATEGORY THEORY PROFESSOR PETER JOHNSTONE Cats have been around for 70 years. Eilenberg + Mac Lane =. Cats are about building bridges between different parts of maths. Definition 1.1. A category C consists
More informationConstructive version of Boolean algebra
Constructive version of Boolean algebra Francesco Ciraulo, Maria Emilia Maietti, Paola Toto Abstract The notion of overlap algebra introduced by G. Sambin provides a constructive version of complete Boolean
More informationCOMPACT ORTHOALGEBRAS
COMPACT ORTHOALGEBRAS ALEXANDER WILCE Abstract. We initiate a study of topological orthoalgebras (TOAs), concentrating on the compact case. Examples of TOAs include topological orthomodular lattices, and
More informationTopological aspects of restriction categories
Calgary 2006, Topological aspects of restriction categories, June 1, 2006 p. 1/22 Topological aspects of restriction categories Robin Cockett robin@cpsc.ucalgary.ca University of Calgary Calgary 2006,
More informationThe Zariski Spectrum of a ring
Thierry Coquand September 2010 Use of prime ideals Let R be a ring. We say that a 0,..., a n is unimodular iff a 0,..., a n = 1 We say that Σa i X i is primitive iff a 0,..., a n is unimodular Theorem:
More informationThe space of located subsets
The space of located subsets Tatsuji Kawai Universtà di Padova Second CORE meeting, 27 January 2017, LMU 1 / 26 The space of located subsets We are interested in a point-free topology on the located subsets
More informationExponentiation of unary topologies over inductively generated formal topologies
Exponentiation of unary topologies over inductively generated formal topologies Maria Emilia Maietti - Silvio Valentini Dipartimento di Matematica Pura ed Applicata Università di Padova via G. Belzoni
More informationMath 249B. Nilpotence of connected solvable groups
Math 249B. Nilpotence of connected solvable groups 1. Motivation and examples In abstract group theory, the descending central series {C i (G)} of a group G is defined recursively by C 0 (G) = G and C
More informationMath 541 Fall 2008 Connectivity Transition from Math 453/503 to Math 541 Ross E. Staffeldt-August 2008
Math 541 Fall 2008 Connectivity Transition from Math 453/503 to Math 541 Ross E. Staffeldt-August 2008 Closed sets We have been operating at a fundamental level at which a topological space is a set together
More informationAlgebraic Geometry Spring 2009
MIT OpenCourseWare http://ocw.mit.edu 18.726 Algebraic Geometry Spring 2009 For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms. 18.726: Algebraic Geometry
More informationarxiv: v1 [math.ct] 28 Dec 2018
arxiv:1812.10941v1 [math.ct] 28 Dec 2018 Janelidze s Categorical Galois Theory as a step in the Joyal and Tierney result Christopher Townsend December 31, 2018 Abstract We show that a trivial case of Janelidze
More informationFormally real local rings, and infinitesimal stability.
Formally real local rings, and infinitesimal stability. Anders Kock We propose here a topos-theoretic substitute for the theory of formally-real field, and real-closed field. By substitute we mean that
More informationABSTRACT DIFFERENTIAL GEOMETRY VIA SHEAF THEORY
ABSTRACT DIFFERENTIAL GEOMETRY VIA SHEAF THEORY ARDA H. DEMIRHAN Abstract. We examine the conditions for uniqueness of differentials in the abstract setting of differential geometry. Then we ll come up
More informationMorita-equivalences for MV-algebras
Morita-equivalences for MV-algebras Olivia Caramello* University of Insubria Geometry and non-classical logics 5-8 September 2017 *Joint work with Anna Carla Russo O. Caramello Morita-equivalences for
More informationCONSTRUCTIVE GELFAND DUALITY FOR C*-ALGEBRAS
CONSTRUCTIVE GELFAND DUALITY FOR C*-ALGEBRAS THIERRY COQUAND COMPUTING SCIENCE DEPARTMENT AT GÖTEBORG UNIVERSITY AND BAS SPITTERS DEPARTMENT OF MATHEMATICS AND COMPUTER SCIENCE, EINDHOVEN UNIVERSITY OF
More information1 Categorical Background
1 Categorical Background 1.1 Categories and Functors Definition 1.1.1 A category C is given by a class of objects, often denoted by ob C, and for any two objects A, B of C a proper set of morphisms C(A,
More informationAlgebraic Geometry Spring 2009
MIT OpenCourseWare http://ocw.mit.edu 18.726 Algebraic Geometry Spring 2009 For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms. 18.726: Algebraic Geometry
More informationUniversal Properties
A categorical look at undergraduate algebra and topology Julia Goedecke Newnham College 24 February 2017, Archimedeans Julia Goedecke (Newnham) 24/02/2017 1 / 30 1 Maths is Abstraction : more abstraction
More informationALGEBRAIC GEOMETRY COURSE NOTES, LECTURE 9: SCHEMES AND THEIR MODULES.
ALGEBRAIC GEOMETRY COURSE NOTES, LECTURE 9: SCHEMES AND THEIR MODULES. ANDREW SALCH 1. Affine schemes. About notation: I am in the habit of writing f (U) instead of f 1 (U) for the preimage of a subset
More informationfive years later Olivia Caramello TACL 2015, 25 June 2015 Université Paris 7 The theory of topos-theoretic bridges, five years later Olivia Caramello
Université Paris 7 TACL 2015, 25 June 2015 2 / 37 unifying in Mathematics In this lecture, whenever I use the word topos, I really mean Grothendieck topos. was introduced in the paper in 2010. The unification
More informationCHAPTER 0: BACKGROUND (SPRING 2009 DRAFT)
CHAPTER 0: BACKGROUND (SPRING 2009 DRAFT) MATH 378, CSUSM. SPRING 2009. AITKEN This chapter reviews some of the background concepts needed for Math 378. This chapter is new to the course (added Spring
More informationOn the normal completion of a Boolean algebra
Journal of Pure and Applied Algebra 181 (2003) 1 14 www.elsevier.com/locate/jpaa On the normal completion of a Boolean algebra B. Banaschewski a, M.M. Ebrahimi b, M. Mahmoudi b; a Department of Mathematics
More informationMath 210B. Profinite group cohomology
Math 210B. Profinite group cohomology 1. Motivation Let {Γ i } be an inverse system of finite groups with surjective transition maps, and define Γ = Γ i equipped with its inverse it topology (i.e., the
More informationFOUNDATIONS OF ALGEBRAIC GEOMETRY CLASS 2
FOUNDATIONS OF ALGEBRAIC GEOMETRY CLASS 2 RAVI VAKIL CONTENTS 1. Where we were 1 2. Yoneda s lemma 2 3. Limits and colimits 6 4. Adjoints 8 First, some bureaucratic details. We will move to 380-F for Monday
More informationPart II. Logic and Set Theory. Year
Part II Year 2018 2017 2016 2015 2014 2013 2012 2011 2010 2009 2008 2007 2006 2005 2018 60 Paper 4, Section II 16G State and prove the ǫ-recursion Theorem. [You may assume the Principle of ǫ- Induction.]
More informationTopology for functional programming
Topology for functional programming Martín Escardó University of Birmingham, England EWSCS, Palmse, Estonia, 26 Feb 2 Mar 2012 Computational fact Function types A B with A infinite don t have decidable
More informationby Say that A is replete (wrt ) if it is orthogonal to all f: X Y in C such that f : Y X is iso, ie for such f and all there is a unique such that the
Electronic Notes in Theoretical Computer Science 6 (1997) URL: http://wwwelseviernl/locate/entcs/volume6html 6pages Studying Repleteness in the Category of Cpos Michael Makkai 1 Department of Mathematics
More informationThe Relation Reflection Scheme
The Relation Reflection Scheme Peter Aczel petera@cs.man.ac.uk Schools of Mathematics and Computer Science The University of Manchester September 14, 2007 1 Introduction In this paper we introduce a new
More informationON THE CONGRUENCE LATTICE OF A FRAME
PACIFIC JOURNAL OF MATHEMATICS Vol. 130, No. 2,1987 ON THE CONGRUENCE LATTICE OF A FRAME B. BANASCHEWSKI, J. L. FRITH AND C. R. A. GILMOUR Recall that the Skula modification SkX of a topological space
More informationThe Adjoint Functor Theorem.
The Adjoint Functor Theorem. Kevin Buzzard February 7, 2012 Last modified 17/06/2002. 1 Introduction. The existence of free groups is immediate from the Adjoint Functor Theorem. Whilst I ve long believed
More informationTopos-theoretic background
opos-theoretic background Olivia Caramello IHÉS September 22, 2014 Contents 1 Introduction 2 2 erminology and notation 3 3 Grothendieck toposes 3 3.1 he notion of site............................ 3 3.2
More information2 logic, such as those for resolution and hyperresolution. A number of recent developments serve as the motivation for the current paper. In [2, 3], i
Sequents, Frames, and Completeness Thierry Coquand 1 and Guo-Qiang Zhang 2?? 1 Department of Computer Science, University of Goteborg S 412 96, Goteborg, Sweden coquand@cs.chalmers.se 2 Department of Computer
More informationALGEBRAIC GROUPS. Disclaimer: There are millions of errors in these notes!
ALGEBRAIC GROUPS Disclaimer: There are millions of errors in these notes! 1. Some algebraic geometry The subject of algebraic groups depends on the interaction between algebraic geometry and group theory.
More information15 Lecture 15: Points and lft maps
15 Lecture 15: Points and lft maps 15.1 A noetherian property Let A be an affinoid algebraic over a non-archimedean field k and X = Spa(A, A 0 ). For any x X, the stalk O X,x is the limit of the directed
More informationCourse 311: Michaelmas Term 2005 Part III: Topics in Commutative Algebra
Course 311: Michaelmas Term 2005 Part III: Topics in Commutative Algebra D. R. Wilkins Contents 3 Topics in Commutative Algebra 2 3.1 Rings and Fields......................... 2 3.2 Ideals...............................
More informationCRITERIA FOR HOMOTOPIC MAPS TO BE SO ALONG MONOTONE HOMOTOPIES
CRITERIA FOR HOMOTOPIC MAPS TO BE SO ALONG MONOTONE HOMOTOPIES SANJEEVI KRISHNAN arxiv:0709.3715v3 [math.at] 5 Dec 2008 Abstract. The state spaces of machines admit the structure of time. A homotopy theory
More informationUniquely Universal Sets
Uniquely Universal Sets 1 Uniquely Universal Sets Abstract 1 Arnold W. Miller We say that X Y satisfies the Uniquely Universal property (UU) iff there exists an open set U X Y such that for every open
More informationDR.RUPNATHJI( DR.RUPAK NATH )
Contents 1 Sets 1 2 The Real Numbers 9 3 Sequences 29 4 Series 59 5 Functions 81 6 Power Series 105 7 The elementary functions 111 Chapter 1 Sets It is very convenient to introduce some notation and terminology
More information3. The Sheaf of Regular Functions
24 Andreas Gathmann 3. The Sheaf of Regular Functions After having defined affine varieties, our next goal must be to say what kind of maps between them we want to consider as morphisms, i. e. as nice
More informationElimination of binary choice sequences
Elimination of binary choice sequences Tatsuji Kawai Japan Advanced Institute of Science and Technology JSPS Core-to-Core Program Workshop on Mathematical Logic and its Application 16 17 September 2016,
More informationVariations on Realizability. Simple examples realizing axioms of choice
Variations on Realizability Simple examples realizing axioms of choice J.M.E. Hyland July 1, 1999 8x 2 X9y 2 Y'(x y)!9f 2 Y X 8x 2 X'(x f (x)) 1 Motivation M.E.Maietti: How close to a topos can one get
More informationConstructive algebra. Thierry Coquand. May 2018
Constructive algebra Thierry Coquand May 2018 Constructive algebra is algebra done in the context of intuitionistic logic 1 H. Lombardi, C. Quitté Commutative Algebra: Constructive Methods, 2016 I. Yengui
More informationMODELS OF HORN THEORIES
MODELS OF HORN THEORIES MICHAEL BARR Abstract. This paper explores the connection between categories of models of Horn theories and models of finite limit theories. The first is a proper subclass of the
More informationarxiv: v1 [math.ct] 28 Oct 2017
BARELY LOCALLY PRESENTABLE CATEGORIES arxiv:1710.10476v1 [math.ct] 28 Oct 2017 L. POSITSELSKI AND J. ROSICKÝ Abstract. We introduce a new class of categories generalizing locally presentable ones. The
More informationFilters in Analysis and Topology
Filters in Analysis and Topology David MacIver July 1, 2004 Abstract The study of filters is a very natural way to talk about convergence in an arbitrary topological space, and carries over nicely into
More informationAffine combinations in affine schemes
Affine combinations in affine schemes Anders Kock Introduction The notion of neighbour points in algebraic geometry is a geometric rendering of the notion of nilpotent elements in commutative rings, and
More informationAmalgamable diagram shapes
Amalgamable diagram shapes Ruiyuan hen Abstract A category has the amalgamation property (AP) if every pushout diagram has a cocone, and the joint embedding property (JEP) if every finite coproduct diagram
More informationThe overlap algebra of regular opens
The overlap algebra of regular opens Francesco Ciraulo Giovanni Sambin Abstract Overlap algebras are complete lattices enriched with an extra primitive relation, called overlap. The new notion of overlap
More informationCategory Theory. Categories. Definition.
Category Theory Category theory is a general mathematical theory of structures, systems of structures and relationships between systems of structures. It provides a unifying and economic mathematical modeling
More informationHandbook of Logic and Proof Techniques for Computer Science
Steven G. Krantz Handbook of Logic and Proof Techniques for Computer Science With 16 Figures BIRKHAUSER SPRINGER BOSTON * NEW YORK Preface xvii 1 Notation and First-Order Logic 1 1.1 The Use of Connectives
More informationA New Category for Semantics
A New Category for Semantics Andrej Bauer and Dana Scott June 2001 Domain theory for denotational semantics is over thirty years old. There are many variations on the idea and many interesting constructs
More informationUniversal Algebra for Logics
Universal Algebra for Logics Joanna GRYGIEL University of Czestochowa Poland j.grygiel@ajd.czest.pl 2005 These notes form Lecture Notes of a short course which I will give at 1st School on Universal Logic
More informationRepresentable presheaves
Representable presheaves March 15, 2017 A presheaf on a category C is a contravariant functor F on C. In particular, for any object X Ob(C) we have the presheaf (of sets) represented by X, that is Hom
More informationTHE SEMICONTINUOUS QUASI-UNIFORMITY OF A FRAME
Pré-Publicações do Departamento de Matemática Universidade de Coimbra Preprint Number 03 27 THE SEMICONTINUOUS QUASI-UNIFORMITY OF A FRAME MARIA JOÃO FERREIRA AND JORGE PICADO Abstract: The aim of this
More information